Nanostructured polystyrene well plates allow unbiased high-throughput characterization of...

Nanostructured Polystyrene Well Plates Allow Unbiased High-Throughput Characterization of Circulating Tumor CellsYuan Wan,† Marnie Winter,† Bahman Delalat,‡ Jennifer E. Hardingham,§,∥ Phulwinder K. Grover,§

Joseph Wrin,§ Nicolas H. Voelcker,‡ Timothy J. Price,§ and Benjamin Thierry*,†

†Ian Wark Research Institute, University of South Australia, Mawson Lakes, Adelaide, South Australia 5095, Australia‡ARC Centre of Excellence in Convergent Bio-Nano Science and Technology, Mawson Institute, University of South Australia,Mawson Lakes, Adelaide, South Australia 5095, Australia§The Queen Elizabeth Hospital and University of Adelaide, Adelaide, South Australia 5011 Australia∥School of Medical Sciences, University of Adelaide, Adelaide, South Australia 5095, Australia

ABSTRACT: Rapid, reliable and unbiased circulating tumor cell (CTC)isolation and molecular characterization methods are urgently requiredfor implementation in routine clinical diagnostic and prognosticprocedures. We report on the development of a novel unbiased CTCdetection approach that combines high-throughput automated micros-copy with a simple yet efficient approach for achieving a high level oftumor cell binding in standard tissue culture polystyrene (PS) wellplates. A single 5 min high-power oxygen plasma treatment was used tocreate homogeneous nanoscale roughness on standard PS tissue cultureplates and, in turn, drastically enhance the binding of a range of tumorcells. After physical adsorption of an adlayer of poly-L-lysine, bindingyields above 97% were obtained at 2 h for all tumor cell lines used in thestudy. Morphological analysis of the cells confirmed strong adherence tothe nanorough PS substrates. Clinically relevant concentrations of ahighly metastatic breast cancer cell line, used as model for CTCs, could be reliably detected among blood cells on the nanoroughpolystyrene plates using an automated microscopy system. The approach was then successfully used to detect CTCs in the bloodof a stage IIIc colorectal cancer patient. By combining the high binding abilities of nanorough PS well plates with the high-throughput nature of high-content analysis systems, this methodology has great potential toward enabling unbiased routineclinical analysis of CTCs. It could be applied, once clinically validated, in any clinical center equipped with an automatedmicroscopy facility at a fraction of the cost of current CTC isolation technologies.KEYWORDS: circulating tumor cells, polystyrene, nanostructure, oxygen plasma, high throughput microscopy, colorectal cancer

1. INTRODUCTIONThe isolation and analysis of circulating tumor cells (CTCs)from cancer patient blood, often referred to as a “liquid biopsy”,can provide important diagnostic and prognostic informa-tion.1−3 For example, enumeration of CTCs is a prognosticatorin patients with early stage and metastatic breast cancer.4

Dynamic changes in CTC loads can also serve as a surrogatebiomarker for the assessment of therapeutic efficacy.5 Inaddition, molecular analysis of CTCs provides a window intothe pathophysiology of metastatic diseases and thus offers thepossibility of monitoring changes in tumor genotypes andguiding anticancer treatment regimens.2,6−8

A number of approaches have been reported for thedetection and isolation of CTCs.9−11 In general, these methodsfall into two categories: biological methods based on affinity-isolation of the target cells and physicochemical methods.Approaches based on affinity interactions provide very highspecificity and efficiency in isolating cells expressing targetedbiomarkers.12 However, because there is currently no

ubiquitous biomarker that can be used to isolate CTCs,biological methods are inherently biased. Of major concern, forinstance, is the occurrence of epithelial to mesenchymaltransition (EMT) in CTCs. CTCs with EMT-like phenotypemay indeed not express the epithelial cell adhesion marker(EpCAM), the transmembrane glycoprotein most widely usedin affinity-based CTCs isolation approaches.13 Low or completelack of EpCAM expression can be expected to result in false-negative findings.14,15 In addition, in certain scenarios CTCsneed to be detached from the isolation platform for further invitro cell culture. Efficient detachment of viable immuno-captured CTCs poses a major postisolation challenge.16

In contrast, physicochemical methods exploit differences insize, deformability, and electrical properties between CTCs andnucleated blood cells. Although isolation purity from physical

Received: August 4, 2014Accepted: November 4, 2014Published: November 4, 2014

Research Article

www.acsami.org

© 2014 American Chemical Society 20828 dx.doi.org/10.1021/am505201s | ACS Appl. Mater. Interfaces 2014, 6, 20828−20836

methods is typically lower than that of biological methods(<10%), they are highly efficient (>90%), label-free, andprovide great flexibility in subsequent enumeration andmolecular profiling of CTCs.10 It is generally assumed thatCTCs are larger than nucleated blood cells, and thisassumption forms the basis of size-based isolation methods;however, CTCs at different stages of apoptosis or in a dormantstate are expected to have smaller size.17 Potential morpho-logical changes associated with EMT also remain to beinvestigated. In addition, the issue of CTC deformability iscontroversial, as opposite observations have been reported.18,19

Thus, the performance in clinical settings of current physicalisolation methods, such as isolation by size of epithelial tumorcells (ISET),20 micropore chip,21 and CellSievo,22 is likely to belower than predicted from model studies with tumor cell lines.Furthermore, despite tremendous advances in this field, bothbiological and physical isolation methods usually remainexpensive and time- and resource-consuming, which limitstheir applicability to clinical uses. Rapid, affordable, and reliableunbiased CTC isolation methods for subsequent enumerationand characterization are still urgently required toward theestablishment of clinically relevant CTC-based prognostic anddiagnostic strategies.As an alternative to enrichment technologies, high-

throughput automated microscopic analysis of the wholenucleated cellular population in a blood sample is attractivebecause it has the potential to provide the required unbiasedsensitive detection of CTCs. Recent developments in high-content imaging platforms enable rapid molecular andmorphological characterization of a large number of cells andcommercial platforms have been already developed for CTCsenumeration.23 To fulfill their clinical potential, such high-content screening approaches will require the development ofsubstrates able to efficiently bind tumor cells present in a bloodsample while also being compatible with automated micros-copy. Nanostructured substrates that mimic the structuralfeatures of the extracellular matrix or basement membrane haverecently been successfully employed in CTCs isolation anddetection.24 Combined with biological ligands, nanostructuredsubstrates provide a high surface area for ligand immobilizationwhile lowering the rolling velocity of cells in microfluidicchannels and, in turn, increasing isolation yields. It has beenrecently reported that tumor cells have very high bindingaffinity on nanostructured surface, even without functionaliza-tion with biological ligands, owing to the dynamic arrangementof integrin-mediated focal adhesions.25 Nanostructured sub-strates are therefore a promising label-free bioplatform for CTCisolation. Due to the limitation of existing nanofabricationmethods, large-scale fabrication of homogeneous nanostruc-tured substrates is, however, nontrivial and also associated withhigh fabrication costs. Thus, the main limitation of reportednanostructure-based CTC isolation techniques is that they cantypically only process a limited amount of blood at a time (<0.5mL). Screening CTCs from high volumes of blood (up to 60mL) would be optimal to provide a true liquid biopsy for solidcancers.26 However, in the clinical setting relevant CTCisolation approaches should be able to process rapidly andefficiently about ∼5−10 mL of blood, especially whenconsidering the small number of CTCs present, especially inearly stage cancer patients.We report here on an integrated CTC detection approach

combining high-throughput total content screening technologywith a simple yet efficient method to significantly increase

tumor cell binding on standard tissue culture polystyrene wellplates. More specifically, we demonstrate that a single 5 minhigh-power oxygen plasma treatment creates homogeneousnanoscale roughness on standard tissue culture polystyrenesubstrates and, in turn, drastically enhances the binding of arange of tumor cells. Oxygen plasma treatment of polystyrene(PS) has been widely used since the 1990s to create optimalsurface for tissue culture.27 Adherent cells can indeed efficientlybind and proliferate on the resulting hydrophilic PS surface.28

This proposed approach enabled rapid, unbiased detection andmolecular analysis of tumor cells spiked in low concentrationsin peripheral blood. Considering the growing availability ofautomated microscopy facilities, nanostructured polystyrenewell plates as described here have a strong potential to providea much-needed technological solution toward large-scaleclinical implementation of CTC-based diagnostic and prog-nostic strategies.

2. MATERIALS AND METHODSAll chemicals were obtained from Sigma-Aldrich, unless otherwisenoted.

2.1. Oxygen Plasma Treatment and Surface Character-ization. Six-well plates (standard Corning Costar 3516) were exposedto oxygen plasma (Plasma asher, Siemens) at 500 W for 5, 10, and 15min at a pressure of 0.1 mbars with a 100 mL/s oxygen flow rate. Tointroduce a dense layer of NH2 functionalities on the surface, freshlyplasma-treated PS wells were incubated with a 0.01% (w/v) poly-L-lysine (PLL) solution (MW: 150, 000−300, 000) in deionized (DI)water for 2 h at RT. After a thorough rinsing with DI water and 1×PBS, the plates were dried with a nitrogen stream.

Surface topography was evaluated quantitatively using a JPK atomicforce microscope (AFM). Changes in surface area and root-mean-square surface roughness (Rq) were measured. Height image ofsamples were captured in ambient air with 15−20% humidity in thetapping model. The analyzed field was 1 × 1 μm, and images wereacquired at a scan rate of 1 Hz with 256 scanning lines. Water contactangles were measured on PS before and after plasma treatment. Adroplet of DI water was placed on the surface of the substrate at roomtemperature (RT), and after 30 s, the contact angle was measuredusing a contact angle goniometer (NRL-100, Rame-Hart). Typically,five measurements were averaged for each run. The surface chemistryof the samples was characterized using X-ray photoelectron spectros-copy (XPS) using a Kratos AXIS Ultra DLD X-ray photoelectronspectrometer with a monochromatic Al Kα X-ray source and ahemispherical analyzer. The pass energy was 20 eV with a resolution of0.3 eV for high-resolution spectra. Spectra were collected at aphotoelectron takeoff angle of 90°. Binding energies were referencedto the C 1s hydrocarbon carbon peak at 285.0 eV to compensate forsurface charging effects.

2.2. Cell Culture. BSR, HEK293, 3T3, SH-SY5Y, T98G, MCF7,MDA-MB231, MDA-MB231 HM, and EMT6 cell lines were culturedin Dulbecco’s modified Eagle’s medium (DMEM) supplemented with10% FBS; penicillin, 100 U/mL; streptomycin, 100 μg/mL; and platedin T-25 tissue culture flasks. Flasks were placed into an incubatormaintained 5% CO2 at 37 °C and 10% humidity. A highly metastaticvariant of the MDA-MB231 breast cancer cell lines (MDA-MB-231-HM) initially reported by Prof. Zhou-Luo Ou (Fudan UniversityShanghai Cancer Center, China) was provided by Dr. Erica Sloan(Monash University). The MDA-MB-231-HM cells were transfectedwith a lentiviral vector for green fluorescent protein (GFP).

2.3. Tumor Cell Binding Yields. To determine the yield ofbinding on the various PS samples, we used a range of tumor cellconcentrations (from 1 × 102 to 5 × 105 cells/mL). Cells suspended inDMEM were added into the treated PS wells, which had been firststerilized using 70% ethanol and incubated for 0.5, 1, 2, 4, and 6 h. Thewells were washed gently to remove unbound cells and the attachedcells were fixed with 4% paraformaldehyde for 1 h at RT. Twentyrepresentative images (×10) were randomly taken for each sample

ACS Applied Materials & Interfaces Research Article

dx.doi.org/10.1021/am505201s | ACS Appl. Mater. Interfaces 2014, 6, 20828−2083620829

with a Nikon TE inverted microscope. Cell numbers per images weremanually counted and the resulting cell densities calculated andexpressed as number of cells per mm2. The binding yield was definedas the number of cells bound to the PS wells divided by the initialseeded cell number. The cell morphology was also recorded, and thepercentage of cells with a flat morphology was calculated as describedpreviously.29,30

2.4. Tumor Cell Isolation Yields from Blood. To accuratelydetermine the isolation yield, we spiked MDA-MB231 HM cells stablytransfected to express GFP into healthy blood. The GFP expressionrate of the MDA-MB231 HM cells was first calculated and wasapproximately 99%. This value was used to correct isolation yields.Then, 200, 400, and 800 μL of 2.5 × 102 cells/mL MDA-MB231 HMcells were spiked in 1 mL of fresh healthy human blood (n = 3); 1 mLof the spiked sample was mixed with 14 mL cold red blood cell (RBC)lysis buffer (containing 15.5 mM NH4Cl, 0.1 mM EDTA, and 1 mMNaHCO3) for 10 min at RT. Nucleated cells, including WBCs, and thespiked cancer cells were washed by centrifugation at 4 °C for 5 min at250g and resuspended in DMEM culture medium. The cellularmixtures were added to the treated PS well plates and incubated for 2h. The typical total cell density on the PS surface was about 1000 cells/mm2. MDA-MB231 HM cells expressing GFP could be readilydetected using fluorescent microscopy. Well plates were analyzedusing the Operetta (PerkinElmer, Waltham, MA) high-contentimaging system using 10× and 20× objectives. MDA-MB231 HMcells isolated in every well were counted, and a linear regressionanalysis between the detected numbers and the spiked numbers wasperformed.Next, to demonstrate the feasibility of immunostaining the isolated

cells, were also used nonfluorescent MCF7. After we isolated the cellsas described above, we stained the cells with an Anti-Human CD45FITC. After a washing step with basic buffer (1× PBS and 0.5%sodium azide), cells were fixed with 4% paraformaldehyde for 1 h atRT and then sequentially treated with permeabilizing block buffer (1×PBS with 0.5% sodium azide supplemented with 0.05% Triton X-100and 10% FBS) and 10 μg/mL of anticytokeratin (AE1/AE3) eFluor660 for 30 min on ice. Cells were then washed with perm wash (basicbuffer supplemented with 0.05% Triton X-100) three times followedby 0.1 μg/mL DAPI staining.2.5. Tumor Cell Retention during Staining. To determine the

loss of tumor cells during immuno-staining, we seeded 2 mL of 2.5 ×104 cells/mL pretreated with the cell lysing buffer, as described beforefor MCF7 cells, into either a PLL-treated nanorough PS well plate orinto an untreated one. Cells were allowed to bind for 2 h and thensubjected to the staining procedure. Cell loss was determined bycounting cell numbers in 8 × 1 mm2 areas randomly selected beforeand after applying the complete staining procedure.2.6. Detection of CTCs in Blood of Colorectal Cancer Patient.

Human ethics approval was obtained from the ethics committee ofThe Queen Elizabeth Hospital. Following approval, 7.5 mL ofperipheral blood was collected from a stage IIIc colorectal cancerpatient. The blood sample was lysed as described above, and the cellswere plated on the roughened plates and then stained with anti-Human CD45, anti-CK, antivimentin eFluor 570, and DAPI.Typically, 6-well plates were analyzed using the Operetta systemusing 10× and 20× objectives.

3. RESULTS AND DISCUSSION3.1. Oxygen Plasma Treatment of PS. PS samples

treated for 15 min were typically deformed as a result of theelevated temperature induced by the high-powered atomicoxygen attack during the plasma treatment. The glass transitiontemperature of PS is approximately 100 °C,31 which limits thetemperatures to which PS samples can be exposed. It was notedthat even for 10 min of treatment, the PS plates occasionallydistorted. The morphology and surface roughness of theoxygen-plasma-treated PS was characterized using AFM. Asshown in Figure 1, homogeneously distributed nanostructured

features were generated on the standard tissue culture PSsubstrates used in this study following high-powered oxygenplasma treatment. Surface features with low aspect ratios andpyramidal shapes with approximately 60 nm lengths of bottomline and 14 nm heights were obtained. The surface roughnessincreased with increasing exposure times. The Rq values ofuntreated PS, 5 min treated PS, and 10 min treated PS were0.85 nm (SD: 0.14 nm), 4.26 nm (SD: 0.23 nm), and 4.49 nm(SD: 0.09 nm), respectively. Considering the only modestincrease in roughness resulted from prolonged treatment withhigh powered oxygen plasma, 5 min plasma treatment wastherefore selected and applied for the rest of the study. XPSelemental analysis showed a significant increase in the oxygenpercentage on the PS samples from 14.2 to 25%, confirmingthat the expected oxidation of the PS surface took place duringthe oxygen plasma (Figure 2A). Next, the water contact anglesof PS samples treated for 5 min with the high-powered oxygenplasma were continuously monitored for 72 h, and the averagevalues and standard deviations were recorded. Static watercontact angle decreased from 52 ± 3° for untreated PS to beundetectably low for oxygen-plasma-treated PS (typicallymeasured within 20 min of the plasma treatment). Contactangles increased to 3 ± 1° at the 24 h time point, 9 ± 1° at the48 h time point, and 11 ± 2° at the 72 h time point. Thesemeasurements show that the high-power oxygen plasmatreatment enables PS to remain superhydrophilic for at least72 h.

3.2. Tumor Cell Binding Yields. Our data demonstratesthat a 5 min oxygen plasma treatment at high powersignificantly increases the roughness of standard tissue culturePS, thereby resulting in homogeneously nanostructured

Figure 1. Representative AFM images (1 × 1 μm) of (A) untreated PSand (B) PS treated with 500 W oxygen plasma for 5 min. The surfaceroughness increased from 0.85 nm for untreated PS to 4.26 nm for thenanostructured plasma treated PS surface.



surfaces. To determine the effect of the nanoscale roughness, 5× 105 T98G (glioblastoma cell line) cells suspended in DMEMwere incubated on various substrates including plasma treatedPS (rough PS), untreated PS (PS), as well as glass used ascontrol for low adhesion for up to 6 h. T98G cell binding yieldsincreased in all three groups with extended incubation time(Figure 3A). At the 2 h time point, for instance, 94 ± 2% of thecells had adhered onto the rough PS while only 73 ± 2.5% and34 ± 4% cells were bound onto the untreated PS and glasssurface, respectively. These results demonstrate that as expectedthe use of nanostructured PS could significantly accelerate thebinding of the T98G tumor cells. The latter likely originate inthe increased roughness and hydrophilicity of the treated PS,offering larger surface area for cell attachment and spreading.28

To further characterize the binding yields, we investigated cellseeding numbers ranging from 102 to 5 × 105 T98G cells permL. An average of 94% binding efficiency at the 2 h time pointwas observed for the rough PS group, showing a reliableconstant sensitivity (Figure 3B). Higher variations wereobserved for the lower seeding concentrations, which wasmainly due to the inherent error associated with manipulationof low cell numbers.Although the binding yield of the T98G cells on rough PS

was increased to 98 ± 3% at the 6 h time point, shortincubation times are preferable toward practical implementa-tion in routine clinical work. In addition, it has been reportedthat a significant percentage of CTCs are in different stages ofapoptosis;32 and CTCs presenting both mild apoptotic changes,

such as cytoplasmic blebbing, and apoptotic indicators in thenucleus have been observed.23 CTCs can be expected to initiatean apoptosis program during longer incubation time in vitroand consequently have reduced binding during the assay.Therefore, rapid attachment of cells onto the substrate ispreferred. On the basis of these considerations and the bindingyield results, we selected a 2 h incubation time as a goodcompromise.Oxygen plasma treatment is a well-established methodology

to modify the surface morphology of polymeric materials andincrease their roughness. For instance, oxygen plasma was usedto introduce nanoscale roughness on polyethylene tereph-thalate and, in turn, drastically influenced the adhesion ofmouse liver cancer cells.33 Tumor cells significantly proliferatedon nanostructured surfaces with a low aspect ratio due toenhanced mechanical anchoring; in contrast, superhydrophilicsurfaces with high aspect ratio nanostructures failed to offersufficient sites for focal adhesion and thus suppressed theadhesion and growth of the mouse liver cancer cells. A templateapproach was recently used to prepare nanostructured PS,which was found to efficiently bind tumor cells.34 Althoughadvanced fabrication technology such as reactive ion etchingcan be used to prepare nanorough surfaces with exquisitecontrol over the morphology of the nanoscale features, oxygenplasma is rapid, cost-efficient, and readily available in mostresearch environments. While further studies are required tofully understand the structure−activity relationships of nano-structured surfaces in regard to cell adhesion, high-poweroxygen plasma treatment of PS is a simple yet efficientapproach to introduce homogeneous nanoscale roughness andfoster tumor cell binding.Next, the binding yields on the nanorough PS of eight

different types of cells, including three of noncancerous celllines (BSR, HEK293 and 3T3), were determined. In agreementwith the data obtained for T98G cells, high binding yields wereobserved for tumor cells (over 91% for all tested cells) on therough PS after 2 h incubation; in contrast, only 76% ofHEK293 cells were bound onto nanorough PS, which isconsistent to the inherently weak adhesion ability of this cellline (Figure 3C). However, for even the HEK293 cells, thebinding yield was significantly enhanced on the rough PSsurface because less than 40% adhered to the untreated tissueculture PS at the 2 h time point. The cellular morphology canreflect the level of cell adhesion. In general, higher contact areabetween cells and substrates, along with flatter morphology(changed from a globular shape to a semielliptical one) withdecreased height is typically linked to strong attachment to thesubstrate.35 On the contrary, a refractile and rounded cellularmorphology indicates relatively weak attachment, which maylead to detachment under high shear stress, for instance duringthe washing steps associated with immunostaining of the cells.To further investigate this aspect, cells adhered to the PSsubstrate were observed in bright field microscopy, and cellswith obvious lamellipodia or filopodia were defined as cells with“flat morphology”. At least 80% of the attached cells wereclassified as flat except for the HEK293 cells (Figure 3D); over90% of the SH-SY5Y (neuroblastoma cell line) and T98G cellsdisplayed flat morphologies.These results demonstrate that the tumor cell lines used here

all had efficient, strong, and rapid attachment onto tissueculture PS treated with 5 min of high-powered oxygen plasma.However, approximately 3−7% of the tested cancer cells didnot attach onto the rough PS at the 2 h time point target. In

Figure 2. (A) XPS elemental analysis of (gray line) untreated and (redline) oxygen-plasma-treated PS samples. (B) XPS elemental analysis ofuntreated PS, PLL adsorbed on untreated PS, and PLL adsorbed onoxygen-plasma-treated PS.



addition, the exact phenotype of CTCs is not yet fullydetermined, and consequently, cancer cell lines cannot fullyrepresent CTCs. Phenotypical changes have indeed beenobserved along with a high level of heterogeneity. The presenceof a significant percentage of apoptotic CTCs is also an issue,although the clinical relevance of these cells remains to beestablished. Toward further improving the binding yield oftumor cells on the rough PS, the addition of an electrostaticlayer based on poly-L-lysine (PLL) resulting in excess positivecharges on the surface was investigated. PLL is commonly usedin cell biology to create positively charged surfaces, whichenhance binding to negatively charged cellular membranes.Freshly oxygen-plasma-treated PS samples were incubated witha PLL aqueous solution and thoroughly washed with PBS. XPSanalysis was used to confirm the presence of the PLL layer ontothe PS. As shown in Figure 2, XPS elemental spectra revealed asignificant increase in the peak at 400 eV attributed to N 1s(6.5%) for PS samples treated for 5 min with oxygen plasmaand incubated with PLL in comparison to untreated PS (2.7%).No significant increase in elemental nitrogen percentage wasdetected after incubating untreated PS with PLL (N 1s = 2.6%).The significant increase in elemental nitrogen on the surfaceconfirmed the adsorption of the polyamino acid on the surface

of the plasma oxygen treated PS. Next, the binding yields andcellular morphology on PLL-treated nanorough PS wasdetermined. For all six types of cancer cells on PLL-coatedrough PS, binding yields were increased in comparison to thenon-PLL-coated samples and reached almost 100% at the 2 htime point (Figure 4). In addition, all the cells adhered to thesurface displayed visible lamellipodia or filopodia.Finally, the binding yield on the PLL-coated rough PS

substrates of pure WBCs collected from 1 mL peripheralhealthy blood was also investigated. Binding yields close to100% were observed, similar to that of cancer cells, at the 2 htime point. This result confirmed that the combination ofexcess positive charges resulting from the PLL adsorbed layerand the nanorough features introduced on the PS by high-powered oxygen plasma is highly efficient to ensure strong andrapid binding of both a wide range of tumor cell lines andWBCs isolated from blood.

3.3. Isolation and Detection of Tumor Cells fromBlood Using Treated PS and Automated Microscopy.Having established a simple and robust protocol to significantlyincrease the binding of tumor cells on standard tissue culturePS well plates, its compatibility with high-throughput detectionof tumor cells from peripheral blood using automated

Figure 3. (A) Binding yields of T98G cells as a function of incubation time on untreated PS, oxygen plasma rough PS, and glass. (B) Binding yieldsat 2 h as a function of seeding density for T98G cells on untreated PS, oxygen plasma rough PS, and glass. (C) Binding yields for different cell lineson untreated PS and oxygen plasma rough PS. For each group, 5 × 105 cells were seeded and incubated at 37 °C for 2 h (*, p < 0.05). (D)Percentage of cells with flat morphology on rough PS substrates. For each group, 5 × 105 cells were seeded and incubated at 37 °C for 2 h.



microscopy was determined. To this end, a high contentanalysis system was used (Operetta, PerkinElmer) to automati-cally screen the PS well plates. Six well plates were first treatedfor 5 min with high-powered oxygen plasma and further reactedwith PLL, as described. A highly metastatic variant of thehuman breast cancer cell lines MDA-MBA231 transfected tostably express the green fluorescent protein was used to modelCTCs (MDA-MB231-HM).36 The MDA-MB231 HM cellswere carefully spiked into blood collected from a healthy donorat known concentrations (50, 100, and 200 cells per mL).The blood samples were subjected to standard lysis,

resuspended in DMEM medium, and finally seeded into thePLL coated rough PS well plates. After 2 h of incubation, thewell plates were washed with PBS and imaged using theOperetta system. The number of fluorescent MDA-MB231-HM cells detected in the wells is shown in Table 1, and

representative images of the cells are shown in Figure 5. Thenumber of detected cells was within the experimental errorinherent to ultralow cell numbers and consistent with thenumber of spiked cells, and a good correlation was calculatedusing a simple regression analysis (R2 = 0.99) demonstratingquasi complete binding of the MDA-MB231-HM cells spiked inthe blood in this assay. With a 20× magnification, themorphology of the attached cancer cells was easily observed,and the presence of lamellipodia evidenced strong binding tothe treated PS substrates. High content screening technologieshave rapidly progressed in recent years, providing powerfulmeans to screen a large number of cells and obtainmultiparametric characterization. Several commercial platforms

have been developed, and our results demonstrate the feasibilityof high-throughput automated isolation and detection of tumorcells from peripheral blood using commercially available wellplates. High-content screening systems such as the Operettahave the potential to provide advanced phenotypical andmorphological characterization, which is a key requirementtoward clinical implementation of CTCs in cancer diagnosticand prognostic applications.Next, to confirm the feasibility of reliably quantification of

tumor cells isolated from peripheral blood and demonstrate thepossibility to carry out phenotypical characterization, we spikednonfluorescent MCF7 cells in blood and isolated using the PLLrough PS approach. Cells were stained in the well plates usinganti CD45 and anticytokeratin (AE1/AE3) antibodies anddetected with the Operetta system, as shown in Figure 6. As

expected, MCF7 cells showed a strong specific cytoplasmiccytokeratin staining pattern, which was not observed on WBCs,thereby enabling their accurate quantification and character-ization.Next, the loss of tumor cells during staining was determined.

MCF-7 cells treated with the lysing buffer were seeded on

Figure 4. Binding yields for different cell lines on PLL rough PS. Foreach group, 5 × 105 cells were seeded and incubated at 37 °C for 2 h.

Table 1. Enumeration of MDA-MB231 HM Cells Spiked inWhole Blood on PLL-Coated Rough PS Using the OperettaSystem

sample

number of spikedcells (based oncalculation)

number of detectedcells (based onfluorescencedetection)

calibratedvalue

detectionrate (%)

1a 50 45 45.5 911b 50 56 56.6 113.21c 50 46 46.5 932a 100 109 110.2 110.22b 100 91 92 912c 100 114 115.3 115.33a 200 207 209.3 104.73b 200 213 215.4 107.73c 200 202 204.3 102.2

Figure 5. MDA-MB231 HM cells were observed among WBCs in a 6-well plate. The panoramagram is composed of approximately 4000images taken at 10× magnification. (Insets) The red box indicates afield of view obtained using the 10× objective lens and containing 2MDA-MB231 HM cells; the white boxes indicate areas ofmagnification obtained using the 20× objective lens and clearlydisplay the morphology of the cells on the PS substrate.

Figure 6. Positive DAPI and anti-CK staining of MCF7 cells isolatedon the rough PS. The picture was taken using the 20× objective lens.



either PLL-treated nanorough PS or in a control tissue culturePS well-plate and allowed to attach for 2 h. The cells were thenwashed and stained as described above. Cell retention on thePLL-treated nanorough PS was found to be 96 ± 3%. On theother hand, decreased cell retention was determined for cellseeded on untreated tissue culture PS (58 ± 17.5%). Theseresults confirm the strong adhesion of tumor cells ontonanorough plasma-treated PS and demonstrate that onlyminimal cell loss occurs during staining, which is an importantfeature when considering very rare events.3.4. Detection of CTCs from Colorectal Cancer

Patient’s Blood Sample. Finally, we aimed to validate acomplete protocol for the detection of CTCs from patients’blood using the proposed rough PS well plates/automatedmicroscopy approach. To this end, an important considerationis the binding capacity of the samples. From our experimentalobservation cell density within the PS wells should not behigher than 3000 cells/mm2 to avoid steric issues and theresulting decrease of cell binding efficiency. With suchconsiderations in mind, 3−4 mL of blood per plate could bereliably analyzed. To validate our approach, we collected bloodfrom a colorectal cancer (stage IIIc) patient. Nucleated cellswere obtained, seeded as described above, and stained andimaged with the Operetta system. To identify CTCs, we usedthe fluorescent markers anti-CD45, anticytokeratin, andantivimentin and coupled them with the cytomorphology asrevealed using the 20× objective. Morphological examinationcan be useful in distinguishing background staining from thetrue CTCs,37 and an inherent advantage of the proposedapproach is the possibility of obtaining high-resolution imagesof the putative CTCs. Cells that showed a positive nuclearstaining with DAPI, a negative staining against CD45, and apositive staining for CK were considered as CTCs (Figure 7).

To rule out false positive WBCs, we strictly excluded cellsshowing clear CD45-positive staining, even if these cellsexhibited abnormal morphology and CK-positive staining.Twenty-one CTCs in the 3.5 mL of blood analyzed wereunambiguously identified. No vimentin positive cells wereobserved. No cells with epithelial phenotype (CK) weredetected in control experiments using blood from healthydonors.

Depending on the number of markers used to characterizethe CTCs and the specifications of the microscopy platform,automated analysis could be completed as quickly as 2 h, givinga total isolation and analysis time of approximately 5 h fromblood collection with minimum manipulations. The proposedapproach is also expected to be compatible with currentpositive and negative CTC enrichment technologies, whichcould further shorten analysis times. For instance, leukapheresiswas recently demonstrated as a clinically safe method to enrichand concentrate CTCs from large blood volume26 and could beeasily integrated into the proposed isolation/detection platformin the future. Finally, a number of commercial automatedmicroscopy platforms are currently available at a fraction of thecost of current commercial CTCs detection systems.

4. CONCLUSIONIn this study, we have demonstrated that a simple 5 min high-power oxygen plasma treatment introduces homogeneouslydistributed nanoscale features on standard tissue culture PSplates. The presence of the nanoscale roughness significantlyincreased the binding of a range of tumor cells, andmorphological analysis demonstrated strong adherence of thecells to the nanorough PS substrates. Physical adsorption of anadlayer of poly-L-lysine resulted in a further modest increase inthe binding yields. At least 97% binding yields were obtained at2 h for all tumor cell lines investigated in the study. Using theOperetta automated high-content analysis system, clinicallyrelevant concentrations of a highly metastatic breast cancer cellline spiked in peripheral blood could be reliably detected on therough polystyrene plates. The proposed approach wassuccessfully used to detect CTCs in the blood of a stage IIIccolorectal cancer patient, demonstrating its feasibility in a real-life situation. The highly efficient tumor cell isolation substratesreported here could be readily used, once clinically validated,on a large scale in any center equipped with automatedmicroscopy facilities. This approach has the potential to providea cost-effective technological solution toward large-scale clinicalimplementation of CTC-based diagnostic and prognosticstrategies

■ AUTHOR INFORMATIONCorresponding Author*E-mail: [email protected] authors declare no competing financial interest.

■ ACKNOWLEDGMENTSThis work was supported by the National Health and MedicalResearch Council (NH&MRC) Project grant APP 631939. Theauthors would like to thank the Australian National FabricationFacility (ANFF) node of South Australia for access to theirfacilities, as well as Prof. Zhou Luo Ou (Fudan University) andDr. Erica Sloan (Monash University) for providing MDA-MB231-HM cells. Thanks also to Dr. Marco Herald at Walterand Eliza Hall Institute for the GFP lentiviral construct used totransfect the MDA-MB231-HM cells.

■ REFERENCES(1) Alix-Panabieres, C.; Pantel, K. Circulating Tumor Cells: LiquidBiopsy of Cancer. Clin. Chem. 2013, 59, 110−118.(2) Kang, Y.; Pantel, K. Tumor Cell Dissemination: EmergingBiological Insights from Animal Models and Cancer Patients. CancerCell 2013, 23, 573−581.

Figure 7. DAPI, CD45, and anticytokeratin staining of CTCs isolatedfrom a colorectal cancer patient. A total of 21 CTCs were detected in3.5 mL of peripheral blood. The picture was taken using the 20×objective lens. BF = bright field. The scale bar equals 10 μm.



(3) van de Stolpe, A.; Pantel, K.; Sleijfer, S.; Terstappen, L. W.; denToonder, J. M. Circulating Tumor Cell Isolation and Diagnostics:Toward Routine Clinical Use. Cancer Res. 2011, 71, 5955−5960.(4) Zhang, L.; Riethdorf, S.; Wu, G.; Wang, T.; Yang, K.; Peng, G.;Liu, J.; Pantel, K. Meta-Analysis of the Prognostic Value of CirculatingTumor Cells in Breast Cancer. Clin. Cancer Res. 2012, 18, 5701−5710.(5) Gorges, T. M.; Pantel, K. Circulating Tumor Cells as Therapy-Related Biomarkers in Cancer Patients. Cancer Immunol. Immunother.2013, 62, 931−939.(6) Joosse, S. A.; Pantel, K. Biologic Challenges in the Detection ofCirculating Tumor Cells. Cancer Res. 2013, 73, 8−11.(7) Yu, M.; Ting, D. T.; Stott, S. L.; Wittner, B. S.; Ozsolak, F.; Paul,S.; Ciciliano, J. C.; Smas, M. E.; Winokur, D.; Gilman, A. J.; Ulman, M.J.; Xega, K.; Contino, G.; Alagesan, B.; Brannigan, B. W.; Milos, P. M.;Ryan, D. P.; Sequist, L. V.; Bardeesy, N.; Ramaswamy, S.; Toner, M.;Maheswaran, S.; Haber, D. A. RNA Sequencing of PancreaticCirculating Tumour Cells Implicates WNT Signalling in Metastasis.Nature 2012, 487, 510−513.(8) Yu, M.; Bardia, A.; Wittner, B. S.; Stott, S. L.; Smas, M. E.; Ting,D. T.; Isakoff, S. J.; Ciciliano, J. C.; Wells, M. N.; Shah, A. M.;Concannon, K. F.; Donaldson, M. C.; Sequist, L. V.; Brachtel, E.;Sgroi, D.; Baselga, J.; Ramaswamy, S.; Toner, M.; Haber, D. A.;Maheswaran, S. Circulating Breast Tumor Cells Exhibit DynamicChanges in Epithelial and Mesenchymal Composition. Science 2013,339, 580−584.(9) Yu, M.; Stott, S.; Toner, M.; Maheswaran, S.; Haber, D. A.Circulating Tumor Cells: Approaches to Isolation and Character-ization. J. Cell Biol. 2011, 192, 373−382.(10) Jin, C.; McFaul, S. M.; Duffy, S. P.; Deng, X.; Tavassoli, P.;Black, P. C.; Ma, H. Technologies for Label-Free Separation ofCirculating Tumor Cells: From Historical Foundations to RecentDevelopments. Lab Chip 2014, 14, 32−44.(11) Hardingham, J. E.; Kotasek, D.; Farmer, B.; Butler, R. N.; Mi, J.-X.; Sage, R. E.; Dobrovic, A. Immunobead-PCR: A Technique for theDetection of Circulating Tumor Cells Using Immunomagnetic Beadsand the Polymerase Chain Reaction. Cancer Res. 1993, 53, 3455−3458.(12) Nagrath, S.; Sequist, L. V.; Maheswaran, S.; Bell, D. W.; Irimia,D.; Ulkus, L.; Smith, M. R.; Kwak, E. L.; Digumarthy, S.; Muzikansky,A.; Ryan, P.; Balis, U. J.; Tompkins, R. G.; Haber, D. A.; Toner, M.Isolation of Rare Circulating Tumour Cells in Cancer Patients byMicrochip Technology. Nature 2007, 450, 1235−1239.(13) Grover, P.; Cummins, A.; Price, T.; Roberts-Thomson, I.;Hardingham, J. Circulating Tumour Cells: The Evolving Concept andthe Inadequacy of Their Enrichment by EpCAM-based Methodologyfor Basic and Clinical Cancer Research. Ann. Oncol. 2014, 25, 1506−1516.(14) Joosse, S. A.; Hannemann, J.; Spotter, J.; Bauche, A.; Andreas,A.; Muller, V.; Pantel, K. Changes in Keratin Expression DuringMetastatic Progression of Breast Cancer: Impact on the Detection ofCirculating Tumor Cells. Clin. Cancer Res. 2012, 18, 993−1003.(15) Mikolajczyk, S. D.; Millar, L. S.; Tsinberg, P.; Coutts, S. M.;Zomorrodi, M.; Pham, T.; Bischoff, F. Z.; Pircher, T. J. Detection ofEpCAM-Negative and Cytokeratin-Negative Circulating Tumor Cellsin Peripheral Blood. J. Oncol. 2011, 2011, 252361.(16) Zheng, Q.; Iqbal, S. M.; Wan, Y. Cell Detachment: Post-Isolation Challenges. Biotechnol. Adv. 2013, 31, 1664−1675.(17) Harouaka, R. A.; Nisic, M.; Zheng, S. Y. Circulating Tumor CellEnrichment Based on Physical Properties. J. Lab. Autom. 2013, 18,455−468.(18) Guo, Q.; Park, S.; Ma, H. Microfluidic Micropipette Aspirationfor Measuring the Deformability of Single Cells. Lab Chip 2012, 12,2687−2695.(19) Zhang, W.; Kai, K.; Choi, D. S.; Iwamoto, T.; Nguyen, Y. H.;Wong, H.; Landis, M. D.; Ueno, N. T.; Chang, J.; Qin, L. MicrofluidicsSeparation Reveals the Stem-Cell-Like Deformability of Tumor-Initiating Cells. Proc. Natl. Acad. Sci. U.S.A. 2012, 109, 18707−18712.(20) Hofman, V. J.; Ilie, M. I.; Bonnetaud, C.; Selva, E.; Long, E.;Molina, T.; Vignaud, J. M.; Flejou, J. F.; Lantuejoul, S.; Piaton, E.;Butori, C.; Mourad, N.; Poudenx, M.; Bahadoran, P.; Sibon, S.;

Guevara, N.; Santini, J.; Venissac, N.; Mouroux, J.; Vielh, P.; Hofman,P. M. Cytopathologic Detection of Circulating Tumor Cells Using theIsolation by Size of Epithelial Tumor Cell Method Promises andPitfalls. Am. J. Clin. Pathol. 2011, 135, 146−156.(21) Asghar, W.; Wan, Y.; Ilyas, A.; Bachoo, R.; Kim, Y. T.; Iqbal, S.M. Electrical Fingerprinting, 3D Profiling, and Detection of TumorCells with Solid-State Micropores. Lab Chip 2012, 12, 2345−2352.(22) Lim, L. S.; Hu, M.; Huang, M. C.; Cheong, W. C.; Gan, A. T.;Looi, X. L.; Leong, S. M.; Koay, E. S.; Li, M. H. Microsieve Lab-ChipDevice for Rapid Enumeration and Fluorescence in Situ Hybridizationof Circulating Tumor Cells. Lab Chip 2012, 12, 4388−4396.(23) Marrinucci, D.; Bethel, K.; Kolatkar, A.; Luttgen, M. S.;Malchiodi, M.; Baehring, F.; Voigt, K.; Lazar, D.; Nieva, J.; Bazhenova,L.; Ko, A. H.; Korn, W. M.; Schram, E.; Coward, M.; Yang, X.;Metzner, T.; Lamy, R.; Honnatti, M.; Yoshioka, C.; Kunken, J.;Petrova, Y.; Sok, D.; Nelson, D.; Kuhn, P. Fluid Biopsy in Patientswith Metastatic Prostate, Pancreatic, and Breast Cancers. Phys. Biol.2012, 9, 016003.(24) Wang, L. X.; Asghar, W.; Demirci, U.; Wan, Y. NanostructuredSubstrates for Isolation of Circulating Tumor Cells. Nano Today 2013,8, 374−387.(25) Chen, W. Q.; Weng, S. N.; Zhang, F.; Allen, S.; Li, X.; Bao, L.W.; Lam, R. H. W.; Macoska, J. A.; Merajver, S. D.; Fu, J. P.Nanoroughened Surfaces for Efficient Capture of Circulating TumorCells without Using Capture Antibodies. ACS Nano 2013, 7, 566−575.(26) Fischer, J. C.; Niederacher, D.; Topp, S. A.; Honisch, E.;Schumacher, S.; Schmitz, N.; Zacarias Fohrding, L.; Vay, C.;Hoffmann, I.; Kasprowicz, N. S.; Hepp, P. G.; Mohrmann, S.; Nitz,U.; Stresemann, A.; Krahn, T.; Henze, T.; Griebsch, E.; Raba, K.; Rox,J. M.; Wenzel, F.; Sproll, C.; Janni, W.; Fehm, T.; Klein, C. A.; Knoefel,W. T.; Stoecklein, N. H. Diagnostic Leukapheresis Enables ReliableDetection of Circulating Tumor Cells of Nonmetastatic CancerPatients. Proc. Natl. Acad. Sci. U.S.A. 2013, 110, 16580−16585.(27) Zeiger, A. S.; Hinton, B.; Van Vliet, K. J. Why the Dish Makes ADifference: Quantitative Comparison of Polystyrene Culture Surfaces.Acta Biomater 2013, 9, 7354−7361.(28) Ishizaki, T.; Saito, N.; Takai, O. Correlation of Cell AdhesiveBehaviors on Superhydrophobic, Superhydrophilic, and Micro-patterned Superhydrophobic/Superhydrophilic Surfaces to TheirSurface Chemistry. Langmuir 2010, 26, 8147−8154.(29) Wan, Y.; Kim, Y. T.; Li, N.; Cho, S. K.; Bachoo, R.; Ellington, A.D.; Iqbal, S. M. Surface-Immobilized Aptamers for Cancer CellIsolation and Microscopic Cytology. Cancer Res. 2010, 70, 9371−9380.(30) Wan, Y.; Mahmood, M. A.; Li, N.; Allen, P. B.; Kim, Y. T.;Bachoo, R.; Ellington, A. D.; Iqbal, S. M. Nanotextured Substrates withImmobilized Aptamers for Cancer Cell Isolation and Cytology. Cancer2012, 118, 1145−1154.(31) Keddie, J. L.; Jones, R. A. L.; Cory, R. A. Size-DependentDepression of the Glass-Transition Temperature in Polymer-Films.Europhys. Lett. 1994, 27, 59−64.(32) Wong, C. W.; Lee, A.; Shientag, L.; Yu, J.; Dong, Y.; Kao, G.; Al-Mehdi, A. B.; Bernhard, E. J.; Muschel, R. J. Apoptosis: An Early Eventin Metastatic Inefficiency. Cancer Res. 2001, 61, 333−338.(33) Ko, T. J.; Kim, E.; Nagashima, S.; Oh, K. H.; Lee, K. R.; Kim, S.;Moon, M. W. Adhesion Behavior of Mouse Liver Cancer Cells onNanostructured Superhydrophobic and Superhydrophilic Surfaces. SoftMatter 2013, 9, 8705−8711.(34) Liu, X.; Chen, L.; Liu, H.; Yang, G.; Zhang, P.; Han, D.; Wang,S.; Jiang, L. Bio-Inspired Soft Polystyrene Nanotube Substrate forRapid and Highly Efficient Breast Cancer-cell Sapture. NPG AsiaMater. 2013, 5, e63.(35) Hallab, N.; Bundy, K.; O’connor, K.; Clark, R.; Moses, R. CellAdhesion to Biomaterials: Correlations Between Surface Charge,Surface Roughness, Adsorbed Protein, and Cell Morphology. J. Long-Term Eff. Med. Implants 1994, 5, 209−231.(36) Chang, X. Z.; Li, D. Q.; Hou, Y. F.; Wu, J.; Lu, J. S.; Di, G. H.;Jin, W.; Ou, Z. L.; Shen, Z. Z.; Shao, Z. M. Identification of theFunctional Role of AF1Q in the Progression of Breast Cancer. BreastCancer Res. Treat. 2008, 111, 65−78.



(37) Williams, A.; Lin, H.; Datar, R.; Cote, R. J. Reporting theCapture Efficiency of A Filter-Based Microdevice: A CTC Is Not aCTC Unless It Is CD45 Negative. Clin. Cancer Res. 2011, 17, 3050.



Nanostructured polystyrene well plates allow unbiased high-throughput characterization of...

Documents

Transcript of Nanostructured polystyrene well plates allow unbiased high-throughput characterization of...